The aim of navigation is to determine the position and the spatial orientation of a moving vehicle. This so-called trajectory can be achieved by using the combination of multiple sensors. The advantages of such a setup is that the benefits of the different systems are used to obtain improved navigation information. A possible multi-sensor system is the integration of GNSS (Global Navigation Satellite Systems) and IMU (Inertial Measurement Unit)), where GNSS provides accurate position information and in contrast, the inertial sensor captures high-dynamic movements and provides a direct orientation determination. At GNSS outages (loss of signal), the navigation information is based on the IMU until the GNSS solution is available again. One of the limiting factors are the temporal increasing sensor errors of the inertial navigation. To achieve a navigation solution in the cm-range, these sensor errors must be modelled. Furthermore, the IMU measurements contain vehicle vibrations and measurement noise beside the actual vehicle motion wanted. Therefore, the resulting position error is proportional to these disorders. The entire multi-sensor navigation solution will most likely be improved when these disorders get eliminated. In this thesis, the wavelet analysis is used to minimize the undesirable effects due to measurement noise and other disturbances. Kalman filtering is used for the sensor integration of GNSS and IMU and in addition, the sensor errors can be estimated within this technique. Finally, the effect of preprocessing the IMU observables is analysed within the multi-sensor navigation system. Thereby improvements of the initial orientation have been achieved, which leads to a smaller position and orientation error at the beginning of the trajectory. Likewise, it was shown that undesirably recorded vibrations can be eliminated, resulting subsequently in a smaller position error at GNSS outages.